ebook img

Is Highly Heterogeneous in Primary Cervical Screening in Amazonian Peru PDF

12 Pages·2015·0.49 MB·English
by  
Save to my drive
Quick download
Download
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview Is Highly Heterogeneous in Primary Cervical Screening in Amazonian Peru

RESEARCHARTICLE Visual Inspection after Acetic Acid (VIA) Is Highly Heterogeneous in Primary Cervical Screening in Amazonian Peru MaribelAlmonte1,2*,CatterinaFerreccio3,SilvanaLuciani4,MiguelGonzales5,Jose M.Delgado5,CarlosSantos6,ManuelAlvarez6,JackCuzick2,PeterSasieni2 1InternationalAgencyforResearchonCancer,Lyon,France,2CentreforCancerPrevention,Wolfson InstituteofPreventiveMedicine,QueenMaryUniversityofLondon,London,UnitedKingdom,3Advanced CenterforChronicDiseases,ACCDIS-FONDAP,FacultaddeMedicina,PontificiaUniversidadCatólicade Chile,Santiago,Chile,4PanAmericanHealthOrganization(PAHO),Washington,DC,UnitedStatesof America,5DirecciónRegionaldeSaludSanMartín,SanMartín,Perú,6InstitutoNacionaldeEnfermedades Neoplásicas,Lima,Perú * [email protected] Abstract OPENACCESS Citation:AlmonteM,FerreccioC,LucianiS, GonzalesM,DelgadoJM,SantosC,etal.(2015) Background VisualInspectionafterAceticAcid(VIA)IsHighly HeterogeneousinPrimaryCervicalScreeningin Conventionalcytology(Pap)andvisualinspectionaftertheapplicationofaceticacid(VIA) AmazonianPeru.PLoSONE10(1):e0115355. arecurrentlyusedinprimaryscreeninginPeru.StudiessuggestthatthequalityofVIAis doi:10.1371/journal.pone.0115355 highlyvariable.Over36000womenwerescreenedwithPapandVIAintheTATI(Tamizaje AcademicEditor:Jian-XinGao,ShanghaiJiaoTong yTratamientoInmediatodeLesionesCervico-uterinas)projectconductedinAmazonian UniversitySchoolofMedicine,CHINA Peru.Withinanestedstudytocompareseveralscreeningtechniques(C-TATI),atotalof Received:June23,2014 5435womenwereadditionallyscreenedwithliquid-basedcytology(LBC)andhigh-risk Accepted:November22,2014 humanpapillomavirustesting(HR-HPV).Weinvestigatethevariationofpositivityratesof VIA,Pap,LBCandHR-HPVinC-TATIandofVIAinthefullTATIintervention. Published:January30,2015 Copyright:©2015Almonteetal.Thisisanopenac- Methods cessarticledistributedunderthetermsofthe CreativeCommonsAttributionLicense,whichpermits Atthescreeningvisit,midwivescollectedthreecervicalsamplesforPap,LBCandHC2be- unrestricteduse,distribution,andreproductioninany foreperformingVIA.Thedispersionfactor“D”(D=Pearsonchi-squarevalue/degrees-of- medium,providedtheoriginalauthorandsourceare freedom)wasusedtomeasurethevariabilityoftestsresults.WithinC-TATI,thevariability credited. ofpositivityratesofVIA,Pap,LBCandHR-HPVwasalsographicallyassessedwithbox- DataAvailabilityStatement:Relevantdataarewith- andscatterplotsbymidwifeandmonthofscreening.Funnelplotsandsmoothedscatter inthepaperanditsSupportingInformationfiles. plotswereusedtocorrelatethevariationofVIAbythenumberofexaminationsperformed Funding:TheTATIprojectwasconductedwiththe byeachmidwifeoverthefullTATIintervention. generoussupportoftheBill&MelindaGatesFounda- tiontothePanAmericanHealthOrganizationthrough theAllianceforCervicalCancerPrevention.Cancer Results ResearchUKprovidedfinancialsupportfortheaddi- ConsistentlyoverTATI,VIAresultswerehighlyvariable,independentlyoftheexaminer,the tionalworkinvolvedinthisstudyincludingthecostof HPVandLBCtesting. timewhenthetestwasperformedandthenumberofteststheexaminerperformed(D>6, p-values<0.001).InC-TATI,VIAresultsvariedthemostwhilethoseofHR-HPVvariedthe CompetingInterests:Theauthorshavedeclared thatnocompetinginterestsexist. least(Ds>25,p-values<0.001forVIA,Ds<1.6,p-values>0.05forHR-HPV).Noevidence PLOSONE|DOI:10.1371/journal.pone.0115355 January30,2015 1/12 VariationofVIAinCervicalScreeninginPeru forcorrelationbetweenthenumberofVIAsdonepermidwifeandthevariabilityofVIAre- sultswasobserved. Conclusion Thelackofover-dispersionforHR-HPVdetectionsuggeststhatthevariableVIAresultsdo notreflecttruevariationinunderlyingdisease,butalackofconsistencyinhuman judgement. Introduction Cervicalcanceristhefourthmostcommonfemalecancerintheworld[1].Cervicalscreening usingconventionalcytology(Pap)hashadanimpactoncervicalcancerratesinmanydevel- opedcountriesbutnotinthedevelopingworld[2,3]. Ithasbeenestablishedthatinfectionwithhigh-risktypesofhumanpapillomavirus(HPV) isanecessarycauseforcervicalcancer[4].Bothhigh-riskHPVdetectiontests(HR-HPV)and prophylacticvaccineshavebeendevelopedandHPVvaccinationprogrammestargetingyoung girlshavebeenestablishedallovertheworld.However,itwilltakeseveraldecadestoseethe impactofHPVvaccination,hence,cervicalscreeningshouldcontinue.HPVtestingwithcytol- ogytriagemayreplacecytologyscreeningsoonwithinorganisedscreeningprogrammeswith good-qualitycytology. Thequestionofhowbesttoscreenpopulationswithdifficultornoaccesstohealthcarere- mainsunanswered.Visualinspectionaftertheapplicationofaceticacid(VIA)followedby treatmentwithcryotherapyhasbeenproposedasanalternativescreeningapproachinthese settings.However,resultsfromstudiesevaluatingtheperformanceofVIAhavebeeninconsis- tent.Inparticular,severalstudiesconductedinIndiaandAfricahavereportedbetterVIAre- sultsthanthoseofLatinAmerica[5–10].TherecentlypublishedWHOguidelinesfor screeningandtreatmentofprecancerouslesionsforcervicalcancerpreventionproposetheuse ofHPVtesting,ifavailable,followedbyimmediatecryotherapyorbytriagewithVIAandthen cryotherapy[11].Itisnowimportantnotonlytoevaluatetheperformanceofscreeningtests butalsotoidentifywhichtestsaremorepronetosubjectivevariation(dependonproviders), whichsufferwhenusedindifficultconditionsinplaceswithfewspecialistdoctorsandpoor communications,andwhichcanbeusedinsimplealgorithmstoensurethatmostwomenwith lesionsreceivedadequatetreatment.Over36000womenwerescreenedwithVIAandPapin the“TamizajeyTratamientoInmediatodeLesionesCervico-uterinas”(TATI)projectinAma- zonianPeru[12].Withinanestedstudythatcomparedseveralscreeningstrategies(C-TATI), 5435womenwereadditionallyscreenedwithliquid-basedcytologyandHR-HPV[7].Here,we investigatethevariationofVIA,Pap,LBCandHR-HPVresultsinC-TATIandofVIAinthe fullTATIintervention. Methods ThemethodsofTATIhavebeendescribedelsewhere[7,12,13].Briefly,36759womenaged 25–49yearsresidentsintheSanMartinregioninPeru,wereexaminedbyamidwife,whotook acervicalsampleforaPapandperformedVIA,whichwasconsideredpositiveifanyaceto- whitelesionwasobservedinorclosetothetransformationzoneandotherwisenegative. WomenpositiveonVIAwereexaminedbyageneraldoctor,whoperformedmagnifiedVIA PLOSONE|DOI:10.1371/journal.pone.0115355 January30,2015 2/12 VariationofVIAinCervicalScreeninginPeru (VIAM),tookbiopsiesandtreatedanymanageablelesionswithcryotherapyorreferred womenforcolposcopyandfurthertreatment. TATIwasconductedover36months.Duringthefirstthreemonths,thelogisticsofthe screeningprocesseswerepiloted(months-3to0).Overthefollowing11months,acomparative study(C-TATI,months1to11)inwhich5435womenwereadditionallyscreenedwithliquid- basecytology(LBC)andhigh-riskHPVtesting(HR-HPV)donebyHybridCaptureII(hc2). SimultaneouslyandupuntiltheendofTATI,womenwerescreenedwithonlyPapandVIA (RestofTATI:R-TATI,months1to33). PapandLBCreportedasASCUSorworse(PapASCUS,LBCASCUS)wereconsidered positiveandthestandardthresholdof1(RLU/co)ofhc2wasusedtoidentifyHR-HPVpositive samples.NoHPVgenotypingwasdoneinthisstudy.Cervicalintraepithelialneoplasiagrade2 orworse(CIN2+)wasthemaindiseaseoutcomeofthestudy. VIAtraining Initialtrainingwasofferedto19midwivesand12generaldoctorsonVIA[7,13].Thetraining includedtheoreticalandpracticalsessionsonperformingVIAandcollectingcervicalsamples forPap,generalaspectsoffemaleanatomy,cervicalcancerprevention,diagnosisandmanage- mentofsexuallytransmittedinfectionsandcommunicationandcounsellingskills.Photo- graphsofcerviceswerereviewedthroughoutthecourse.TraineespracticedVIAfirston anatomicalmodelsandthenexaminedwomenunderdirectsupervisionoftrainers.General doctorswereadditionallytrainedonVIAMandtreatmentwithcryotherapy.Atotalof42mid- wivesand25doctorsweretrainedinTATIoverthreetrainingcoursesofoneweekeach. Statisticalanalysis Weusedthedispersionfactor(D)tomeasurethevariabilityofthescreeningtestsresultsbya numberoffactors:age(infive-yearbands),midwifewhoperformedVIAandmonthofscreen- ing,overallandseparatelybysub-study(C-TATIvs.R-TATI).Dwasdefinedastheratioof thePearsonchi-squarevalueofa2xmtable(2=positiveornegative,m=numberoflevelsof factorofinterest)dividedbythedegreesoffreedom(d.f.=n=m-1).Sincetheexpectedvalue ofarandomvariableXwithachi-squaredistributionwith“n”d.f.is“n”,theexpectedvalueof Dunderthenullhypothesisofequalunderlyingpropertiesineachstrataisapproximately1, i.e.E(D)=E(X)/n=1.Thus,Dvaluesclosetozerowillrepresentminimaldispersion(lessthan randomvariation)andhighestvaluesofD,furtheraboveunity,willrepresentoverdispersion. Positivityrates(PR)ofVIA,PapASCUS,LBCASCUSandHR-HPVwerecalculatedover- all,byage,bymonthofscreeningandbymidwifewhocollectedthesamplesandperformed VIA.ThePRofeachmidwife(42differentVIAPRs)participatinginTATIwascalculatedin- cludingtheresultsofallexaminationsperformedthroughouttheinterventionandgrouped intoGroup1:VIAperformedby19midwiveswhoparticipatedinC-TATI(19VIAPRs)and Group2:VIAperformedby23midwiveswhoparticipatedinR-TATIbutnotinC-TATI (23VIAPRs).Asthe19C-TATImidwivesinGroup1screenedwomeninC-TATIandin R-TATIinparallel(inalternativescheduledclinics),theirPRswerealsocalculatedseparately forexaminationsdoneinC-TATI(19PRs)andR-TATI(19PRs).Next,theoverallPRwithin eachgroupwascalculatedsumminguptheresultsofallVIAsperformedinthegroupwithcor- respondingbinomial95%confidenceintervals(95%CI).Atestfortrendswasusedtocompare VIAPRsovertime(year1=months1–11,year2:12–23,andyear3=24–33)inR-TATI. WegraphicallyassessedthevariationofPRswith:1)boxplotstocomparethevariationof PRsofscreeningtestsinC-TATI,2)error-barplotstosummarisethedifferencebetweenVIA PRsperformedbythesamemidwifeduringthefirstyearofinterventioninC-TATIandR-TATI, PLOSONE|DOI:10.1371/journal.pone.0115355 January30,2015 3/12 VariationofVIAinCervicalScreeninginPeru 3)funnelplotstoevaluatetheextenttowhichthevariabilityofVIAresultscouldbeexplainedby thenumberofexaminationsperformedbyeachmidwifeand4)smoothedscatterplotstoevalu- atewhetherthevariabilityofVIAresultscouldbeexplainedbyalearningcurveandifsotoiden- tifyaminimalnumberofVIAexaminationsrequiredtoobtainlessvariableresults[14]. BothfunnelplotsandsmoothedscatterplotsallowvisualisationofthevariationofVIAPRs takingintoaccountthenumberofVIAsperformed.Funnelplotsweregraphedintwosteps: 1)scatterplotsofVIAPRsbythenumberofVIAsperformed,and2)funnelswith95%CIs (controllimits)wereconstructedaroundtheoverallVIAPR.Inthesegraphs,theareaswithin thecontrollimitsrepresenttheamountofvariationexpected.Thus,thehigherthenumberof PRsoutsidethecontrollimits,themoreover-dispersedthedataare;andthemostcloselyclus- teredthePRsaretotheoverallrate,thelessrandomvariation.Arunninglinewasusedto smooththescatterplotsforeachmidwifeandobtainthesmoothedscatterplots. AllstatisticalanalysesweredoneusingtheStatasoftware,version12. EthicalConsiderations TheTATIprojectwasapprovedbytheEthicalCommitteeoftheMinistryofHealthofPeru andtheEthicsCommitteeofthePanAmericanHealthOrganization.Writteninformedcon- sentwasobtainedfromallparticipantsbytrainedmidwives. Results Afterexcluding812VIAsperformedbydoctors,atotalof35947VIAswereperformedby42 midwivesover36months.Duringthepilotphaseoftheintervention,2108womenwere screenedwithPapandVIA.Overthenext11months,19midwivesparticipatinginC-TATI performed12515VIAexaminations,5401withinC-TATIand7114withinR-TATI;while568 VIAswereperformedbynineofthe23midwiveswhoonlyparticipatedinR-TATI.Once C-TATIfinalised,VIAwasperformedin20756women(61%oftheseexaminationsweredone bythe19C-TATImidwives).Fromhereon,resultsarebasedon33839womenwhowere screenedbyamidwifeafterthepilotphase(Fig.1). InC-TATI(n=5401),theoverallPRswere24.2%forVIA,2.1%forPapASCUS+,17.9% forLBCASCUS+and12.6%forHR-HPVtesting.Eachofthesetestswerepositivein36(43%), 32(39%),64(77%)and79(95%)ofthe83CIN2+detectedcases.ThePRsofVIAand HR-HPVtestingdecreasedwithincreasingagewhilethoseforPapandLBCincreasedwith increasingage,mostlyduetoanincreasingtrendofmoderatedysplasiaandworseabnormali- tiesbyage(allp-valuesfortrendswerehighlysignificant,S1Table).Fig.2showsboxplotsof thevariationofPRsofscreeningtestsbymidwifeandovertime.VIAPRsvariedthemost (D=27.8and27.7bymidwifeandovertime,p-values<0.001).Incontrast,HR-HPVtesting wasthemosthomogeneoustest(D=1.5and0.7,p-values>0.05,respectively).PRsofLBC ASCUS+alsovariedsignificantly(D=3.1and8.7,p-values<0.001)whilethoseofPapASCUS +variedlittle(D=1.4and0.7,p-values>0.05).Aftertakingintoaccountthelowrateofdisease detectedbyPap(S1Table),werepeatedtheanalysisoninadequaterates(IR)ofbothLBCand Pap.ThistimethePapIRswerelargelymoreheterogeneous(D=29.0bymidwifeand12.2 overtime,p-values<0.001)thanLBCIRsandPRsofLBCandHR-HPV(datanotshown).We alsorepeatedtheanalysisonpositivityratesofVIAandPapinR-TATI.Theresultsweresimi- lar,VIAPRsweremoreheterogeneousthanPapsmearPRs,however,theresultsofbothtests werehighlyvariable(datanotshown). AsummaryofVIAPRspermidwifebygroupofmidwivesispresentedinTable1.Overall, midwivesperformedbetween30and2346VIAs.Twomidwivesperformedlessthan100VIA exams(onedid30,theother80),43performedbetween113and1962VIAseachandthe PLOSONE|DOI:10.1371/journal.pone.0115355 January30,2015 4/12 VariationofVIAinCervicalScreeninginPeru Figure1.DiagramoftheTATIprojectwiththenumberofvisualexaminationsperformedbymidwives inthepilotphase(months-3–0),duringthetimeofthecomparativestudy(months1–11)andafterthe comparativestudy(months12–33).VIA=Visualinspectionaftertheapplicationofaceticacid.C-TATI= Comparativescreeningstudy(months1–11).R-TATI=restoftheTATIintervention(months1–33). doi:10.1371/journal.pone.0115355.g001 remainingthreeover2000VIAs(2174,2256and2346,respectively).TheoverallVIAPRper midwifeinGroup1was16%(95%CI:15.5%,16.5%)andwassignificantlylowerthanthatof VIAperformedbymidwivesinGroup2(23.5%,95%CI:22.6%,24.4%).Incontrast,thedisper- sionfactorofVIAresultsofmidwivesinGroup1wasalmostdoublethatofmidwivesin Group2(D=48.2andD=27.0,p-values<0.001),althoughbothdispersionfactorswerehighly significant.Theresultsdidnotchangeafterexcludingmidwiveswhoperformedlessthan100 ormorethan2000VIAexaminations.Surprisingly,thePRofmidwivesinGroup1wassignifi- cantlyhigherwhentheyscreenedwomeninC-TATI(24.2%,95%CI:23.1%,25.4%)thanwhen theyscreenedwomeninR-TATI(13.7%,95%CI:13.3%,14.2%),andhigherthantheoverall PRofallmidwivestogether(17.9,95%CI:17.5%,18.3%).Inviewoftheseresults,wecompared PRsinC-TATIandR-TATIofthe19midwivesinGroup1byplottingthedifferencebetween thetwopositivityrates(C-TATI—R-TATI)with95%CIs.Thedifferencewasnegative (decreaseof0.4%)onlyforonemidwifewhiletheresthadhigherPRsinC-TATIthanin R-TATI(meandifference:13.6%,95%CI:-10.9%,38.2%).Thus,midwivesassignedmorefre- quentlyapositiveresultwithinC-TATIthanwithinR-TATI(S1Fig.). Table2showstheVIAPRsofmidwivesbygroupandyearoftheintervention,whenthey screenedwomenonlyinR-TATI.VIAPRsofmidwivesinGroup1wereconsistentlylower thanthoseofmidwivesinGroup2(atleast7%differencebetweenoverallrates,statistically PLOSONE|DOI:10.1371/journal.pone.0115355 January30,2015 5/12 VariationofVIAinCervicalScreeninginPeru Figure2.BoxplotsofpositivityratesofVIA,HPV,LBCandPapandofinadequateratesofLBCand PapinC-TATI.VIA=Visualinspectionaftertheapplicationofaceticacid.HPV=Humanpapillomavirus. LBC=Liquid-basedcytology.C-TATI=Comparativescreeningstudy(months1–11).D=Dispersionfactor. ASCUS+=Atypicalsquamouschangesofundeterminedsignificanceorworsecytologicalabnormalities. Inad=Inadequate. doi:10.1371/journal.pone.0115355.g002 significanteachyear).TherewasaconsistentincreasingtrendoftheoverallPRspergroup ofmidwivesandinbothgroupscombinedfromthefirstyeartothesecondyear,andaconsis- tentsmalldecreasefromthesecondtothethirdyear(p-for-trend<0.001).However,allDs wereonceagainhighlysignificant.ItisworthnotingthatfivemidwivesinGroup2performed lessthan50VIAsinthefirstyear,hencetherangeofVIAPRsvariedbetween0%and67% (D=8.6).Afterexcludingextremevalues,theresultsdidnotchange(range:8.6%to39.5%, D=12.5,p-value<0.001). Insummary,independentlyofwhether:i)VIAwasperformedinC-TATIorinR-TATI, ii)VIAwasperformedbymidwivesinGroup1orGroup2,oriii)VIAwasperformedinthe PLOSONE|DOI:10.1371/journal.pone.0115355 January30,2015 6/12 VariationofVIAinCervicalScreeninginPeru Table1.VariationofVIApositivityratespermidwifeoverthe33monthsoftheTATIinterventionbygroupofmidwives. Groupofmidwifes(No.) No.womenscreened No.VIA/midwife %VIApositive/midwife Dispersionfactor1 Mean Range Overall(95%CI) Range Group1(19) C-TATI 5401 284.3 122–544 24.2(23.1–25.4) 6.8–58.2 27.8 R-TATI 19,748 1039.4 357–2063 13.7(13.3–14.2) 7.0–24.4 30.9 Allgroup1 25,149 1323.6 638–2346 16.0(15.5–16.5) 7.2–27.2 48.2 Group2(23) R-TATI 8690 377.8 30–983 23.5(22.6–24.4) 7.1–43.4 27.0 Allmidwives(42) 33,839 554.7 30–2346 17.9(17.5–18.3) 7.1–43.4 43.0 C-TATI=Comparativescreeningstudy(months1–11).R-TATI=restoftheTATIintervention(months1–33).Group1:19midwiveswhoparticipatedin C-TATIandR-TATI.Group2:23midwiveswhodidnotparticipateinC-TATI.95%CI=95%confidenceinterval. 1Alldispersionfactorswerehighlysignificant(p-value<0.001) doi:10.1371/journal.pone.0115355.t001 first,secondorthirdyear;VIApositivityratesweresignificantlyandconsistentlyheteroge- neous(D>7,p-values<0.001forallpossiblecombinations). Fig.3ashowsthevariationofVIAPRsofeachofthe19Group1midwives(denotedbycap- italletters“A”to“S”)whentheyscreenedwomeninC-TATI(left,ingreydiamonds)andinR- TATI(right,inblackdiamonds).WithinC-TATI,fiveVIAPRswereabove35%(B,C,E,L andN)andthreebelow10%(F,RandS)whileinR-TATIallrateswerebelow25%withsixof thembelow10%(F,J,L,P,RandS).DespitetheapparentcorrelationofVIApositivitywith thenumberofexamsdone(asmidwivesperformedmoreVIAs,theirPRsbecamemoretightly clusteredaroundtheoverallVIAPR);themajorityofVIAPRsliedoutsidethe95%control limitsandonlysix(A,G,J,K,MandO)inC-TATIandfive(A,E,H,MandO)inR-TATI werewithintheexpectedrange.Therewasasubstantial43%reductionintheoverallVIAPR withinR-TATIincomparisontoC-TATI,therefore;wedecidedtorepeatthisanalysis,includ- ingallmidwivesperformingVIAonlyinR-TATI(Fig.3b).Thistime,Group2VIAPRs(de- notedbysmallletters“a”to“w”)wereplottedingreysquaresandthoseofGroup1(denoted byletters“A”to“S”)inblackdiamondsasinFig.3a.ThepatternofVIAPRsperformedinR- TATIbymidwivesinGroup2whodidnotparticipateinC-TATI,wassomehowsimilarto thatofVIAPRsinC-TATIshowninFig.3a.Thistime,theratesof13ofthe23midwivesin Group2werelowerthan20%;however,onlyfivepositivityrateswerewithintheexpected range(a,c,n,sandw). Table2.VariationofVIApositivityratespermidwifebygroupofmidwivesandyearofinterventionwithinR-TATI. GroupofMidwives Year1February-December Year2January-December Year3January-September No.VIAa Mean %VIApositive/midwife Db No.VIAa %VIApositive/midwife Db No.VIAa %VIApositive/midwife Db (No.Mid.) No.VIA Overall(95%CI) (No.Mid.) Overall(95%CI) (No.Mid.) Overall(95%CI) Group1R-TATI 7114(19) 7114(19) 12.2(11.5–12.0) 7.4 6933(19) 15.0(14.2–15.8) 12.5 5701(18) 14.1(13.2–15.0) 17.5 Group2R-TATI 568(9) 568(9) 21.1(17.8–24.7) 8.6 3032(19) 27.1(25.6–28.8) 13.7 5090(21) 21.5(20.4–22.7) 20.7 All 7682(28) 7682(28) 12.9(2.2–13.7) 9.9 9965(38) 18.7(17.9–19.5) 19.3 10,791(39) 17.6(16.9–18.3) 21.9 R-TATI=restoftheTATIintervention(months1–33).Group1:19midwiveswhoparticipatedinC-TATIandR-TATI.Group2:23midwiveswhodidnot participateinC-TATI.Mid.=Midwives.D=Dispersionfactor.95%CI=95%confidenceinterval. aOverallnumberofVIAexaminationsperformedbymidwivesinthatgroupduringthatyear bAlldispersionfactorswerehighlystatisticallysignificant(p-value<0.001) doi:10.1371/journal.pone.0115355.t002 PLOSONE|DOI:10.1371/journal.pone.0115355 January30,2015 7/12 VariationofVIAinCervicalScreeninginPeru Figure3.FunnelplotsofthevariationofpositivityratesofVIAperformedbymidwivesover33 monthsoftheTATIintervention.C-TATI=Comparativescreeningstudy(months1–11).R-TATI=restof theTATIintervention(months1–33).Group1:19midwiveswhoparticipatedinC-TATIandR-TATI.Group 2:23midwiveswhodidnotparticipateinC-TATI.PositivityratesofVIAperformedbymidwivesinGroup1 arerepresentedbycapitalletters“A”to“S”.PositivityratesofVIAperformedbymidwivesinGroup2are representedbysmallletters“a”to“w”. doi:10.1371/journal.pone.0115355.g003 PLOSONE|DOI:10.1371/journal.pone.0115355 January30,2015 8/12 VariationofVIAinCervicalScreeninginPeru Figure4.ProportionofVIAexamsreportedpositiveaccordingtothenumberofVIAexamscarried outbymidwiveswhoperformed500ormoreVIAexamsthroughouttheintervention.ProportionofVIA examsreportedpositivebymidwivesinGroup1arerepresentedbycapitallettersandthoseofmidwivesin Group2bysmallletters. doi:10.1371/journal.pone.0115355.g004 Fig.4showstheproportionofVIAsreportedpositiveovertime(i.e.accordingtothenum- berofVIAscarriedout)foreachofthe25midwiveswhoperformedatleast500VIAs.Forin- stancemidwifeAwasinitiallycallingveryfewVIAspositive.Bythetimeshehasperformed 500VIAsshewascallingout20%positive.Theproportionpositivefellsharplytoabout8% after1000examinationsandthenroseslowlytoabout19%with2000examinations.Beyond 2000theproportionfellbackslightlytoabout15%.Thedifferentpatternsofthesmoothed curvesconfirmedthevariationofVIAresultsandsuggestedthatthisvariabilitycouldnotbe explainedbythenumberofVIAsperformed.Noevidenceforalearningcurveoraminimal numberofVIAsneededtoachieveproficiencywasobserved. Discussion WehaveclearlydemonstratedthatmostofthevariationinVIAPRsinTATImaybeattributed tolackofconsistencyininterpretation.Therewassubstantialvariabilityintheproportionof PLOSONE|DOI:10.1371/journal.pone.0115355 January30,2015 9/12 VariationofVIAinCervicalScreeninginPeru examinationscalledpositivebymonthandbymidwifewhoperformedthetestinC-TATI;in contrastHR-HPVtestingresultswereconsistentlyhomogeneous.Theseresultssuggestedthat thevariationinVIAPRswasnotduetodiseasevariation(1.6%overallrateofCIN2+,range: 1.0–4.4,D=0.20)buttothesubjectivenatureofthetest. Surprisingly,despitethefactthatthe19C-TATImidwivesscreenedwomeninC-TATIand inR-TATIinparallel,theoverallPRofVIAperformedbytheminC-TATI(24.2%)wastwice thePRwhentheyscreenedwomeninR-TATI(13.7%).Indeed,18of19VIAPRsinC-TATI werehigherthaninR-TATIduringthefirst11monthsoftheintervention.Onepotentialex- planationforthesedifferencescouldbebleedingofthecervixassociatedtoadditionalsampling (LBCandHPVafterPap)inC-TATI.Thebleedingandthehighprevalenceofvaginalinfec- tionsmayhaveinterferedwiththevisualisationofthetransformationzone,affectingVIAper- formance,despitethis,VIAresultsinC-TATIweresimilartothoseofotherstudies[7,15–19]. TheminimalnumberofVIAsandtheaveragemonthlyrateofVIAsthataproviderneeds toachieveVIAproficiencyhavenotyetbeendetermined[2].Onfirstinspection,ouranalysis suggestedthatasthenumberofVIAsperformedincreases,theexaminergetsmoreexpertise onthetest,isabletobetterdistinguishaceto-whitelesionswithhigherlikelihoodofbeingtrue lesions,andgrantslessfalsepositiveresults(Fig.3).Thiswasreinforcedbysimilarpatternsof variationamongVIAresultsofthe19C-TATImidwivesinC-TATI(Fig.3a)andthoseofthe 23R-TATImidwivesinR-TATI(Fig.3b).However,whenweplottedtheproportionofVIA positiveresultsofmidwiveswhoatleastperformed500exams,therewerenoconsistentpat- ternsofVIAPRs,contradictingourpreviousobservationsandthatofalearningcurveofVIA afterinitialtraining[19]. Similarlytopreviousstudies,thehighvariabilityofVIAresultspermidwifewasnotex- plainedbylackofadequateanduniformtraining[18,20–23].Inourstudy,midwiveswerein- tensivelytrainedoveraweekinwhichtheyperformedVIAunderrigoroussupervision. Furthermore,wheneverwomenhadhigh-gradecytologyandnegativeVIA,theywerere- examinedbythemidwifeandanexpertgynaecologisttogetherandreasonsfornotvisualising lesionswerediscussed. Ourstudyhassomelimitations.First,wewerenotabletocontrastpositivityrateswithdis- easeascertainmentinR-TATIbecauseonlyPapandVIAwereusedtoreferwomenandun- doubtedlyseveralcasesofCIN2+weremissed.However,weperformedthiscomparison withinC-TATIandweshowedthatthehighvariationofVIAresultswasnotduetodisease variationbuttohumanjudgement.Second,midwivesstartedperformingVIAatdifferenttime pointsinTATIandonlyfourmidwivesperformedVIAeachmonthofTATI,possiblyaffecting thelikelihoodofachievingproficiency.However,weevaluatedthevariationofVIAresultsper- formedinamonthpermidwifeamongmidwiveswhoatleastperformed500examinations, andfoundnoconsistentpatternamongtheproportionofVIApositiveresultsofmidwives. Clearly,thestrongestevidencewouldhavecomefromhavingmultiplemidwivesexamining eachwomanblindedtoeachother’sopinionorbyassigningwomentomidwivesatrandom. Neitherofsuchoptionswaspractical,andsowehavereliedonanalysisofdatathatcouldbe subjecttoconfounding.Nevertheless,ourstudyisthefirsttocomparetheVIAresultsof42 midwiveswhoaltogetherperformedmorethan36000VIAswithintheirroutineclinicalactivi- tiesinalow-resourcesetting,andtheresultsarerobusttothemethodofanalysisandarelikely toreflecttruevariation. Despiteinconsistentresultsonitsperformance,VIAisusuallythemainscreeningtechnique inuseinlow-resourcesettings,partlybecauseitistheonlysimplelow-costtestavailableand partlybecauseVIAresultsareknownimmediatelyallowingsee-and-treatschemes.NewHR- HPVlow-costtestsmaysoonbecomeavailableandHPV-and-treatorHPV-VIA-and-treat schemesarealreadybeingconsideredaspotentialalternativestoVIA-and-cryotherapy.The PLOSONE|DOI:10.1371/journal.pone.0115355 January30,2015 10/12

Description:
The PR of each midwife (42 different VIA PRs) participating in TATI was calculated .. Vedantham H, Silver MI, Kalpana B, Rekha C, Karuna BP, et al.
See more

The list of books you might like

Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.